Shop Forum More Submit  Join Login
For more than a year I didn't have any inspiration. Actually I think my last two submissions were even pretty boring. Now I have many new ideas again. I have at least 5 more renders planned right now, deeper and with more iterations than ever before.

I have so much to say related to my latest image submission "Trees revisited" that I decided to write a journal about it. I hope to clarify what I mean by the term evolution. Information about what it is is spread out over comment sections and deviation descriptions. Also there's been a breakthrough in computer-assisted zooming, which is what's helping me to zoom this deep.

This is "Trees Revisited":
Trees Revisited by DinkydauSet

Evolution zoom method

Maybe the title of "Trees Revisited" is misleading because it's not really about the trees. It's the same old trees again. Instead this is a variation of what I have come to call the evolution zoom method. In general, given some shape that lies somewhere in the Mandelbrot set, evolution can be described as:
1. double the shape
2. morph one of the two copies
3. repeat by treating the morphing of step 2 as the new shape

julia_morphing.png (1024×533)
In the left image, 2 points are labeled 1 and 2 respectively. Zooming in on the point labeled 1, which is outside of the "shape" yields the middle image, a doubling. Zooming in on the point labeled 2 yields the right image, a morphing.

Doubling a shape can be done by zooming to 3/4 (as a good rule of thumb - it's a little more complex than this) of the depth of a minibrot of choice outside of the shape. The exact result depends on the choice of the minibrot. A doubling leads to two copies of the same shape next to each other. That's step one. Morphing one of them involves choosing a minibrot INSIDE that shape, so we choose one, but that means it's not inside the second copy of the shape, so the second copy gets doubled, causing both the morphed shape and two copies of the original shape to be present in the result, which is a set of shapes. By iterating the steps, the original shape and every morphing tied to an iteration of the steps are present in the result and all visible at once. That allows one to see how the original shape evolved, iteration by iteration of the steps, into the final morphing. That's why I call the result an evolution set.

Here's what's new: So at each iteration of the steps we have a morphing and two copies of the previous stage. The way I used to do step 1 in pretty much every previous render where I mentioned the word "evolution" was to morph one of those two copies, but I realized many other ways could be used to double. The only requirement is that the chosen minibrot is outside of the shape to be doubled. I tried a few things and this is the most interesting one I was able to find, at least thus far.

Automated zooming

There is also a lot to be said about the computer assisted zooming I have used to get to this shape. Claude on found an algorithm to determine the location and depth of the nearest minibrot inside a bounded region, involving the Newton-Raphson method. Because doubling and morphing shapes is equivalent to choosing a minibrot and zooming to 3/4 of the depth, knowing where the minibrot is and how deep it is allows one to find the coordinate and the depth of the morphing immediately. The coordinate is the same. The depth (the exponent in the magnification factor required for the minibrot to fill the screen) needs to be multiplied by 3/4. All you need to do is do a few zooms manually to make sure the algorithm searches for the correct minibrot and the computer can do the rest. Kalles Fraktaler has this algorithm implemented and I've been using it a lot. Some links to information about how it works can be found here:…

This is revolutionary. I think we can call it the best invention since the perturbation and series approximation thing. Zooming manually takes A LOT of time. I have spent days to several weeks just zooming for one image. Once the desired path has been chosen, it's a very simple and boring process of zooming in on the center until the required depth is reached. Note that this is not what the algorithm does. It doesn't need to render any pixels or use any visual reference whatsoever. It's a solid mathematics-based method and it works if you give it an "accurate enough" guess of where the minibrot is. Note also that it doesn't help in choosing a location to zoom to. You really just tell it "zoom into this center" and it finds the minibrot inside it for you, saving a lot of work.

It's pretty fast generally, usually faster than manual zooming, especially in locations with few iterations. Based on my experience with the Newton-Raphson zooming in Kalles Fraktaler thus far, I think it's actually a lot slower than manual zooming for locations with a high iteration count. Usually that's still more than made up for. You can work, sleep, study and (most importantly, of course) explore other parts of the mandelbrot set while the computer works for you, 24/7. If you have a processor with many cores you can let it zoom to several locations at once. Effectively that makes it faster in almost every situation.

The evolution zoom method involves a number of iterations of a few steps and I have found that generally it holds that the more steps taken, the better the result. The way the result looks like converges to a limit as the number of steps goes to infinity. The Newton-Raphson zooming allows me to perform more such iterations without as much effort as before. I always want to push the limits of what's possible, so I will perform those extra iterations, meaning I will be zooming a lot deeper. It will lead to shapes that are even more refined with even more symmetries and patterns.

Perturbation for the Mandelbrot set

Perturbation for rendering the Mandelbrot set has been around for a while. I would have written a journal before because it's very awesome, but right from the start there was a fundamental problem: reliability. A recent discovery by Pauldelbrot on indicates that perturbation can now be used to render the Mandelbrot set reliably. Is the project approaching completion? "Correctness" now appears to be achieved.


Roughly a year ago, Kevin Martin published a relatively short document about the Mandelbrot set, containing some equations that staggered everyone. His idea was to apply the principle of perturbation to rendering the Mandelbrot set, and combining that with something he called series approximation. Perturbation allows the iteration count of a pixel to be derived from a different, fully calculated pixel "nearby" (to be called a reference pixel). In practice this means that it's possible to calculate just one single pixel in an image, and derive the rest using perturbation. At great depths with millions of iterations, this saves an enormous amount of render time, which is the main result.

Series approximation allows large number of iterations of pixels to be skipped entirely, good for another enormous speed-up, but it doesn't stop there. In addition, no arbitrary precision calculations are required to do the "deriving" work. Floating point calculations, which are much faster to perform, are sufficient. Martin concludes his document with the following statement:
Using [the equations] means that the time taken rendering Mandelbrot images is largely independent of depth and iteration count, and mainly depends on the complexity of the image being created.
The implications of this are enormous and such a theory is of course yelling to be implemented. Along with the mathematics, Martin also published a simple software implementation of the theory dubbed SuperFractalThing, so that everyone could see that it works. Since then, more software developers have started working on their own implementations.

The simple equation of the Mandelbrot set has long been famous of being so computationally intensive that it can bring any supercomputer to it's knees, as long as you zoom in deep enough. Although that is still the case even with perturbation, the barrier has been shifted significantly. To get an idea of the speed-up we're talking about, consider the following deviation:
SSSSSurvival of the fittest - Evolution #3 by DinkydauSet
Fractal extreme has been the fastest software to calculate the Mandelbrot set for a long time, using traditional optimizations. If the deviation above were to be rendered in Fractal extreme, the render would take roughly 6 months. The actual image was rendered in 6 hours using an implementation of perturbation by Botond Kosa. What you're looking at right there is something that, without perturbation, would have been totally out of reach for many years, no matter how optimized the software is. As Bruce Dawson, the man behind Fractal extreme, commented on good algorithms beat optimized code.


Although there is no doubt that perturbation is a "good algorithm", it came with severe problems right from the start, that Kevin Martin couldn't solve himself. If you have been paying attention, you may have noticed the requirement of a reference pixel to be "nearby". More specifically, usage of floating point numbers to do the calculations requires some numbers in the perturbation equation to be "small". Mathematically, this is completely useless, because there's no exact definition of what "small" is. Indeed, the results of the calculations were shown to be unreliable in many cases. It turned out that the results were correct "most of the time", but sometimes not. Incorrect parts of renders have since been called glitches.

An example of such a glitch can be seen in the image below.
rare_glitch.png (1024×533)
Look closely at the largest spirals. The render on the left contains glitches; the render on the right is correct.

Several attempts have been done to get rid of these inaccuracies. There have been made workarounds where the computer was taught what glitches usually look like, so that they can be automatically recognized and solved. A way to do it is to calculate a second reference point inside the glitched area and do the perturbation calculations again. Having a new reference point more "nearby" solves the glitch. Karl Runmo made notable contributions to this automated glitch solving in his software implementation called Kalles Fraktaler.

As you may understand, it is very difficult to teach a computer to distinguish between correct and incorrect renders visually, especially because glitches can occur in such an enormous variety of types. Even fractal structures can sometimes appear as glitches, which is interesting on its own, but very, very difficult to auto-recognize. As such, manually solving glitches appeared to be a necessity: a very time-consuming process.

It might seem reasonable to spend some time to solve the glitches. Considering how many months of render time (and hundreds of euros worth of electricity) can be saved, spending a day solving glitches doesn't seem so bad. This idea slowly started to change as more difficult types of glitches were found where the "extra-reference-trick" didn't even work. Where does it stop? How many more types of glitches are there and can there ever be made workarounds for all of them? What was needed was more insight in where the inaccuracies come from, so that they can be avoided instead of worked around.

Correctness: now achieved?

Recently, Pauldelbrot on published an algorithm to find reliably which pixels of a render are correct and which aren't. This information can then be used to reliably solve the glitches as well. This was somewhat unexpected, because the algorithm doesn't help in preventing glitches, instead, it helps to detect them afterwards. This is somewhat similar to the approach of Karl Runmo, except Pauldelbrot detects glitches in a non-visual way. The algorithm has shown to be reliable. It automatically solves all the hard-to-detect glitches and no counterexample that slips trough has been found so far. That is great news!

This doesn't mean the project is really finished. There may still be a better way to get rid of glitches still to be discovered and many of the programs that currently use perturbation are still under development. It may even be possible to extend the method of perturbation to be used with different fractals. A good first candidate would be the Mandelbrot set with a power of 3 (instead of 2), but applications in 3d fractal rendering cannot be excluded in the future. The search continues. Mathematics never ends.

Applications in art

I haven't been sitting idle as the developments went on. As such I can now present to you a new video. I once remarked on YouTube that I could do so many more interesting things if just my computer was 1000 times faster. Here you have it. This is one of the things I was thinking of at the time.

More is coming "soon".
The zoom video that I mainly purchased my computer for is here. After 11,5 months of rendering, here is probably the most calculation-intensive Mandelbrot zoom to date.
If you prefer constant zoom speed:…

Render time for this video was 11,5 months, almost a year! I did render other stuff too while this was rendering, but I'm sure it has gotten way over half the CPU time.

Download original file at 60 fps and 50 Mb/s:!s1MX2YTJ!ILjLg0wf…

6.763271516386836374713460138655 E914

And then something else: I saw that there's a HTML edit mode for journals, but it automatically linkifies URLs, automatically adds new lines (no need for br) and it doesn't accept youtube embed code. What's the purpose of HTML edit mode then!?
Real life fractal animation in the wok! Jupiter storms alike…

This effect is a combination of oil and soap which shows very well how the water flows in the pan. It looks very similar to a fractal animation with waves and gnarls.
Damn it! Why can't apophysis ever finish a render that takes a while? I have the patience to wait a few days for a large render to finish, but not to lose the progress every fucking time! There's something terribly wrong with that renderer. As soon as render time passes a few days, it will just stop functioning out of nothing. This is extremely frustrating and preventing me from posting anything. Should I just, instead, render at lower resolutions and have grain all over the image? I've tried apophysis 7x, but it accepts way less RAM-usage than 2.08 3d. I would try the latest 64-bit 7x by Xyrus, but it doesn't allow plugins, which renders it useless for practically every render (pun intended).